Solutions > Manage VMs with recommendations > Best practices for recommendations

Best practices for recommendations

Recommendations provide intelligent resolutions for active and predictive issues in your environment. Depending on the issue and resolution steps, recommendations may be inter-linked or affect each other after every resolution and polling. For example, allocation warnings and CPU contention could have related underlying issues.

To better apply and manage recommendations:

  • Apply active recommendations quickly as possible. The NOW flag indicates the issue has occurred in the environment. These issues actively affect users.
  • Use alerts until recommendations have collected enough data to effectively resolve issues. Active recommendations require 1 hour of data. Predictive recommendations require a minimum of 7 days of data, providing better resolutions and trend tracking with 4+ weeks of data.
  • Select power cycle steps when moving VMs. Hot swapping a VM could cause potential issues if under heavy, constant load during business hours.
  • Schedule virtual system moves during maintenance hours. Moving VMs may require power cycling the VM or interrupting service. Complete these moves during off-peak hours.
  • Monitor environment performance before issuing additional recommendations. Recommendations may affect multiple clusters, hosts, datastores, and other recommendations when allocating resources and moving VMs. Monitor your environment and recommendations as changes and polling completes.
  • Review alerts with active recommendations. If available, recommendations provide links to related alerts. Predictive recommendation may have related potential virtualization alerts, without a displayed link.

Recommendations trigger based on:

  • Alerts that occurred on a cluster, host, datastore, network, or VM in your virtual environment

    Not all alerts have triggered recommendations. If a recommendation has an associated alert, it is linked in the recommendation.

  • VM resource allocations and performance trends and data that project potential issues
  • Enabled strategies for recommendation settings: VM right sizing optimizations, host performance and capacity assurance, storage capacity assurance, and balancing VM workloads on hosts
  • Configured virtualization threshold settings for VMs

If not enough data has been collected, a message displays on the page:

Recommendations do not trigger for VMs with assigned exclusion policies.